Posted 24 May 1999
Interviewed by Peter Nelson
ARF: Your work, although concentrating upon pathological states, seems
applicable to both normal and disease-related biology. Is there a particular
hypothesis which drives your work?
KD: When you look at an AD brain at post-mortem, the most obvious
neuropathological features are the presence of plaques and tangles, but it is
not obvious which of these lesions causes the damage, or whether what you are
looking at is the toxic lesion or something that forms in response to it. In
1991, the discovery of pathogenic mutations in APP that affect APP processing
and the generation of plaque-associated Aβ strongly linked Aβ elevation with
disease etiology. We, and many other labs, therefore developed transgenic mice
with APP transgenes to address the simple hypothesis that elevating Aβ levels
through the overexpression of mutant or wild type human APP would lead to
Alzheimer-type pathology. Early attempts along those lines failed to generate
mice with robust amyloid pathology however, most likely because we now know that
one has to achieve extremely high levels of APP/Aβ for a mouse to develop
amyloid deposits. The idea that Aβ and pathology development were linked was
further justified by the discovery of mutations in the presenilins, where the
effect of pathogenic mutations is to elevate Aβ42 levels specifically. So far,
all the genetic risk factors that have been identified can influence APP or Aβ
so we still feel that modulating amyloid, or its precursor Aβ, is a valid
approach to modeling AD.
ARF: So, the transgenic mouse in Alzheimer's research was an idea
before its time, but its time has come?
KD: To some extent. Despite the initial problems in achieving
sufficient levels of Aβ in the brain, the approach has held up and several mouse
lines now have very robust amyloid pathology. Having thought that this would be
sufficient to generate an animal model with full-blown AD though, we were
surprised to find that the other prominent features of Alzheimer's disease—
overt cell loss and the development of tangles—did not follow. So it would
seem that we've achieved some of the pathology of AD but not all, although many
of the missing features may be present, but more subtle, in the mouse than in
ARF: How much do you think that those observations detract from the
KD: Initially I was disturbed that mice with so much amyloid did not
show more features of the human disease, but the more closely we look at our
transgenic models, the more we see features that do resemble Alzheimer's. For
example, a subtle feature in amyloid depositing transgenic mice that is similar
to AD is that increasing amyloid load is associated with disruption of the
cholinergic system (Wong et al., 1999). In addition, the evidence that amyloid or Aβ accumulation is
associated with cognitive impairment as proposed early on by Karen Hsiao and
Paul Chapman, is also gaining more credibility as several of the amyloid models
now report behavioral changes or electrophysiological disturbances.
Unfortunately the field is still confounded to some extent by the lack of
standardized test paradigms and different genetic backgrounds, so there is still
contradictory data, but overall, many of the models are showing some form of
impairment linked to increasing amyloid/Aβ load. The absence of severe cell loss
in mice with large amounts of amyloid is still puzzling, but there are several
possible explanations ranging from differences in mouse neurobiology to problems
in quantifying cell or tissue loss, shrinkage or dysfunction. To address this,
my colleagues Joe Helpern and Craig Branch at NKI have been developing protocols
for functional and structural MRI which have been informative in identifying how
the transgenic mouse brain responds to amyloid load. The great advantage of MRI
is that live mice can be studied longitudinally, and their disease progression
can be mapped, which we hope will have utility in establishing correlates
between events that can be seen by immunohistochemistry or behavioral analysis
for example, and those that can be seen by imaging.
ARF: Your paper (Duff et al., 1996) provided important data linking presenilin mutation and
beta-amyloid peptide increases in vivo. Do you think that the beta-amyloid
peptide is the lynch pin molecule in the progression of the disease? Are you open
to other explanations?
KD: Although I still believe that Aβ or amyloid accumulation is
critical for the initiation of AD because of the genetic data, I think that
several other factors are likely to be important in the progression of the
disease. In the human AD brain, it is very likely that pathogenic tau
contributes significantly to the degenerative phenotype, and the identification
of mutations in tau that are linked to other neurodegenerative diseases greatly
strengthens that speculation. The role of the inflammatory system is also likely
to be very important, including the potential role of microglia in the
generation or clearance of amyloid from the brain. Genetic analysis may yet help
us to identify susceptibility loci in genes involved in these types of pathways,
which may not be directly linked to Aβ or amyloid accumulation, but to the
brain's ability to deal with it.
ARF: You mentioned some of the alternative hypotheses concerning
Alzheimer's disease. It seems you have been involved in many of the areas of
Alzheimer's research, including ApoE, beta-APP, and presenilins. And more
recently you've done work in tau proteins. Could you talk about that?
KD: Because the amyloid mice are now, albeit in their own way, showing
so many of the features of human AD, I think the next big challenge is to
generate tangles. Because there are differences in the forms of tau present in
the mouse and human brain, we have humanized the tau environment of the mouse by
generating transgenics that express all the forms of human tau. This type of
mouse is currently being crossed to mice that make amyloid deposits and we are
hoping that the resulting offspring will have both amyloid and tau pathology so
we can examine the relationship between these two important features of the
disease. This may help answer the age-old question of what comes first, plaques
or tangles and help us address how each contributes to the AD phenotype.
ARF: Can you describe the differences between your tau transgenic
mouse and the Goedert (Gotz et al., 1995) one?
KD: The transgenic mice that we have created carry a genomic transgene
that includes the human tau promoter and all the coding and regulatory elements
of the human tau gene within a PAC vector. When introduced into the mouse by
microinjection, the human tau protein is represented in the brain with the
correct spatial and temporal distribution. So we've set the scene for
replicating the conditions expected in the human AD brain. The difference
between the genomic tau mouse and others is that most only express a single
isoform of human tau, which is not under the control of the tau promoter. So
they don't get the exact same distribution you'd expect in the human brain, nor
the ability for several human isoforms to aggregate as they do in AD tangles.
ARF: Do you have any preliminary data?
KD: Not yet for the crosses, but in collaboration with Peter Davies we
have characterized the tau mice and have shown that they are making all the
expected isoforms of human tau, and that the protein is expressed at high levels
and with the correct distribution. There is four or five times more human tau in
the mouse brain than mouse tau, and the levels are comparable to what one finds
in the human brain. In addition, the human tau is able to be phosphorylated at
sites that are thought to be significant in AD.
ARF: Wow. Well, on the topic of "patterns," how do you explain the
neuroanatomical specificity of the progression of Alzheimer's disease?
KD: This can be a difficult question for people working with
transgenics to address as transgene derived proteins under the control of a
heterologous promoter are often not distributed at the same levels or in the
same pattern as the endogenous protein. Interpreting regional differences in,
for example, plaque development can therefore be misleading. Having said that,
several of the APP transgenic mice do seem to demonstrate some of the
neuroanatomical correlates that one sees in human Alzheimer's cases. For
example, in both the transgenic mouse and the AD brain, deposits develop early
in the entorhinal cortex, but are rarely seen in the cerebellum, even though Aβ
is found in both regions. There seem to be some areas of the brain that are
particularly vulnerable to Aβ accumulation that is difficult to explain given
our current knowledge. It may be informative to see what local factors are
associated with those areas and in this respect, the regional presence of
proteases, inhibitors or chaperones may be significant, or it could be that more
general mechanisms such as blood flow or clearance are less well able to prevent
Aβ or amyloid from accumulating in sensitive brain regions.
ARF: Regarding genetics, you have done important work both in
transgenics and also in hard-core basic genetics. How do you think the road
ahead shapes up for genetics in Alzheimer's research?
KD: There's still a lot of genetic analysis being done but it's work
with a diminishing return as it takes more work to find fewer and fewer genes,
and the associations are becoming less clear-cut as the genetics becomes more
complicated. As in the past, it is likely that most of the new genetic findings
will identify proteins that can be linked to the amyloid/Aβ pathway, albeit by
six degrees of separation in some cases, but it is likely that other susceptible
pathways will be identified as amyloid load and degree of dementia do not always
correlate. A convincing genetic association between tau and AD is still missing
however, which weakens the argument that tau itself can initiate, or modulate
ARF: Dr. Selkoe and others have touted the gamma-secretase as a
possible therapy target. Do you think there'll be a good therapy option soon in
KD: If, as the genetic data suggests, Aβ elevation or accumulation
truly is the cause of AD, and amyloid modulation is the appropriate therapeutic
goal, then we should be close to achieving that goal.
ARF: Yes! It seems as though transgenic mice, including those which
you developed, provide an extraordinary asset. Have you had much such
KD: The mice that I have developed are very useful in that respect
because they have such a robust and predictable amyloid phenotype. This makes
them very useful for looking at the effects of amyloid modulation not only
because the results can be obtained quickly, but because relatively few mice are
required for statistical significance, and subtle changes in Aβ can be seen more
easily. Now we are at the stage where excellent models have been created, and
drugs have been developed, but often we cannot get the two things together
because of commercial and legal restrictions imposed both by academic
institutions and industry.
ARF: Do you see roadblocks to the development of therapies?
KD: Yes, they have already significantly slowed down progress in the
AD field. Academia has changed a lot even in the past five years and in several
areas, it is tied up by legal considerations. Especially in the case of
transgenic mice, you can't transfer animal resources freely between labs without
a material transfer agreement, and sometimes the terms of the agreement,
especially reach-through rights, make it impossible for two institutions, even
academic institutions, to interact. This has been particularly evident with the
Mayo-owned APP mice. Unfortunately, once one institution acts bullishly, other
institutions often follow suit. It negatively impacts science and wastes money
as resources have to be re-created.
ARF: Can you give an example?
KD: Yes, I get endless requests from academics for the doubly
transgenic PS/APP mice, especially for pilot studies for grants. Unfortunately,
the mice have to be obtained separately from two institutions, accompanied by
legal paperwork and often a lengthy delay, then they must be crossed, genotyped
and aged at the requestee's institute before they can be used for their intended
proof of concept experiment. Often it is not worth the time and resources and
potentially important studies do not get done, all because I cannot send out a
couple of mice to other academics. It's a real shame.
ARF: The unfortunate thing is that the closer we get to findings of
therapeutic value, the slower the progress will be and that's the time when the
most speed seems merited.
KD: For drug companies, there are huge financial, legal and downstream
royalty considerations to deal with when licensing the mice and it often takes
years before they can get their drugs into the model, even for pilot studies.
Putting a drug candidate into a transgenic model is often the last step between
drug development and clinical trials and for some, it's a decisive step in
determining whether the drug will be clinically relevant and should be taken
further. Although it doesn't affect the process of taking the drug to market--I
don't think that FDA approval depends on using a transgenic model-- it's a great
bonus if it shows utility in a model. Certainly, companies pay a lot of money to
have mouse resources for preliminary studies, and when there is money involved
it affects the way people behave, even in 'not-for-profit' foundations.
ARF: It seems as though it also undermines some important foundations
of academia--interaction and collaboration.
KD: Yes, it does, I have experienced it first-hand, both in my career
and my research, and I have seen radical changes in the field in the past five
ARF: What a conflict. It seems complicated, and I hope it gets
KD: It is beginning to be addressed. I recently participated in an NIH
panel on transgenic models and we made several suggestions as to how the models
should be maintained and distributed in the future. Although it is very
complicated because many academics have industrial links, or wish to use
proprietary treatments in an academic setting, guidelines for improved access to
NIH funded animals and less restrictive legal documents were proposed.
Unfortunately though, even the question of who owns a transgenic line can be
very complicated, especially if the animal carries a mutation as patents filed
to cover a mutation will try to also cover mice created using that mutation.
ARF: ...So it's gone beyond transgenics...
KD: Yes, The genetics field has always been particularly prone to
commercial and legal influences. If you discover a mutation then a whole battery
of lawyers will get involved. The mutation may be used for diagnostic purposes,
the development of transgenic mice and potential drug discovery...the whole
painful legal process usually starts with the identification of a genetic
ARF: So it used to be, the first thing you'd do upon discovering a
mutation is write a paper, and now the first thing you'd do is call a
KD: I wouldn't be surprised.
ARF: Maybe it is an outcome that is related to success in the field;
growing pains, or something.
KD: Yes, it is a predictable outcome of high-impact science, and one
that we cannot turn back from, but as a field, we need to impose some rules on
ourselves so that the most aggressive and self-serving individuals don't
determine the future outcome of our efforts.
ARF: Perhaps this is a good segue, then. Do you have any advice to
younger investigators in this shifting field?
KD: Hm, yes: remain interactive. Try to be open-minded, generous and
as open and honest as possible and beware of conflict of interest. Conflict of
interest is very destructive, but what constitutes it is often a personal
judgment that most people won't agree on.
ARF: Okay, and is there anything special to younger female
KD: Getting started with a female-friendly mentor in a healthy lab
environment can be very important. If you're going for an interview, its usually
not that difficult to spot a good lab - there should be a reasonable number of
women in the lab in general but more significantly, a couple at the higher
levels. In general, I think women in science need to be more confident, more
bold in discussions as junior women scientists still tend to be shy in academic
forums, and, often through their own reticence, do not gain credit for their own
work and ideas. Overall, I have not had any discrimination problems that I
couldn't overcome, and although having a social life is almost impossible, I
enjoy being a scientist.
ARF: Well, I think we've taken enough of your time. Thanks very much,
Dr. Duff, for being interviewed for our web site. We truly appreciate your time,
in addition to the important insights and refreshing candor.
Biographical Notes on Karen Duff
- 1987 B.Sc: University of East Anglia, UK
- 1991 Ph.D: University of Cambridge, UK, molecular genetic analysis of
chromosome 21-linked cardiac malformations.
- 1991-1992 Post doc (Dr Alison Goate) St. Mary's Hospital Medical
School/University of London UK
- 1992-1993 Post doc (Dr John Hardy) University of South Florida, Tampa
- 1993-1996 Assistant Professor, University of South Florida, Tampa
(biochemistry and psychiatry departments)
- 1996-1998 Associate Consultant/professor, Mayo clinic Jacksonville, FL
- 1998-present Associate Professor Nathan Kline Institute/NYU medical school
Wong TP, Debeir T, Duff K, Cuello AC. Reorganization of cholinergic terminals in the cerebral cortex and hippocampus in transgenic mice carrying mutated presenilin-1 and amyloid precursor protein transgenes. J Neurosci. 1999 Apr 1;19(7):2706-16. Abstract
Begley JG, Duan W, Chan S, Duff K, Mattson MP. Altered calcium homeostasis and mitochondrial dysfunction in cortical synaptic compartments of presenilin-1 mutant mice. J Neurochem. 1999 Mar;72(3):1030-9. Abstract
Duff K. Alzheimer transgenic mouse models come of age. Trends Neurosci. 1997 Jul;20(7):279-80. Review. No abstract available.
Duff K, Eckman C, Zehr C, Yu X, Prada CM, Perez-Tur J, Hutton M, Buee L, Harigaya Y, Yager D, Morgan D, Gordon MN, Holcomb L, Refolo L, Zenk B, Hardy J, Younkin S. Increased amyloid-beta42(43) in brains of mice expressing mutant presenilin 1. Nature. 1996 Oct 24;383(6602):710-3. Abstract